It is well established in cognitive psychology that stress impairs the retrieval of stored information, often leading to the phenomenon commonly known as a 'blackout' during high-stakes oral examinations. Thus, this mixed-methods study investigated how peer-led and lecturer-led oral examination simulations impact physiological stress responses and students’ self-perceptions. Using heart rate variability (HRV) markers (RMSSD, pNN50, SDNN, LF/HF), we assessed autonomic activation during the simulations. Linear Mixed Models (LMMs) revealed a significant effect of the exam condition on HRV parameters (RMSSD: β = -14.16, p < .001; LF/HF: β = 19.52, p < .001), indicating reduced parasympathetic activation in both simulation types. However, lecturer-led simulations elicited a stronger sympathetic shift (LF/HF interaction: β = -19.33, p < .001), suggesting a higher stress response than peer-led simulations.

Students' competence and confidence significantly improved post-simulation (lecturer-led: t = -8.41, p < .001; student-led: t = -5.82, p < .001), while test anxiety significantly decreased after participating in the simulation (p < .05). Qualitative follow-up data supported the perceived value of simulations for exam preparation and provided insights into the underlying emotional dynamics.

These findings suggest that peer-led simulations may offer a scalable, resource-efficient alternative to lecturer-led training. The positive results highlight the potential of integrating peer-led simulations longitudinally and encourage further research on their cumulative effects over multiple exam cycles.